Lab: Fine-Tune a Binary Classification Model

Lab: Fine-Tune a Binary Classification Model

Use Case: Medical Fraud

About this Mission

On this mission, you’ll learn how an imbalanced dataset can affect a binary classification model’s performance, in this case, a model that predicts the likelihood of medical fraud. You’ll optimize the model based on the estimated cost of the fraud versus the cost of auditing a provider. You will also investigate how a rules-based approach can segment the data using a RuleFit classifier model.

Mission format and duration: self-paced, hands-on, 1 hour

Upon completion of this mission, you will be able to: 

  • Build and evaluate a binary classification model that predicts medical fraud
  • Perform a profit-loss analysis using the Profit Curve tool to optimize the predictive threshold of your model
  • Use the Hot Spots tool with a RuleFit classifier model to understand how a rules-based approach would segment the data
  • Use the Word Cloud tool to understand the relationship between the medications prescribed and the likelihood of a provider perpetrating fraud
  • Make predictions based on the model

Who should complete this mission?

  • Business Analysts
  • Citizen Data Scientists
  • Data Scientists

Before embarking on this mission, you should complete one of the following:

Technical requirements

  • Chrome browser
  • DataRobot Automated Machine Learning — If you don’t have access to the application, please sign up for our free trial: datarobot.com/trial.

 

Action Items1 hr

  • Complete the Lab
  • Give Us Your Feedback
  • Resources
  • Datasets

About this Mission

On this mission, you’ll learn how an imbalanced dataset can affect a binary classification model’s performance, in this case, a model that predicts the likelihood of medical fraud. You’ll optimize the model based on the estimated cost of the fraud versus the cost of auditing a provider. You will also investigate how a rules-based approach can segment the data using a RuleFit classifier model.

Mission format and duration: self-paced, hands-on, 1 hour

Upon completion of this mission, you will be able to: 

  • Build and evaluate a binary classification model that predicts medical fraud
  • Perform a profit-loss analysis using the Profit Curve tool to optimize the predictive threshold of your model
  • Use the Hot Spots tool with a RuleFit classifier model to understand how a rules-based approach would segment the data
  • Use the Word Cloud tool to understand the relationship between the medications prescribed and the likelihood of a provider perpetrating fraud
  • Make predictions based on the model

Who should complete this mission?

  • Business Analysts
  • Citizen Data Scientists
  • Data Scientists

Before embarking on this mission, you should complete one of the following:

Technical requirements

  • Chrome browser
  • DataRobot Automated Machine Learning — If you don’t have access to the application, please sign up for our free trial: datarobot.com/trial.

 

Action Items1 hr

  • Complete the Lab
  • Give Us Your Feedback
  • Resources
  • Datasets